This article was downloaded by: [Sabanci University]

On: 13 August 2015, At: 01:11

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: 5 Howick Place,
London, SW1P 1WG

Electric Power Components and Systems

Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/uemp20

A Dual-loop Model Predictive Voltage Control/Slidingmode Current Control for Voltage Source Inverter
Operation in Smart Microgrids
Khaled M. Abo-al-ez

a b

, Ahmed Elaiw

b c

& Xiaohua Xia

Department of Electrical Engineering, Faculty of Engineering , Mansoura University ,

Mansoura , Egypt
b

Centre of New Energy Systems, Department of Electrical, Electronic and Computer

Engineering , University of Pretoria , South Africa
c

Department of Mathematics, Faculty of Science , King Abdulaziz University , Jeddah , Saudi

Arabia
Published online: 05 Feb 2014.

Click for updates

To cite this article: Khaled M. Abo-al-ez , Ahmed Elaiw & Xiaohua Xia (2014) A Dual-loop Model Predictive Voltage Control/
Sliding-mode Current Control for Voltage Source Inverter Operation in Smart Microgrids, Electric Power Components and
Systems, 42:3-4, 348-360, DOI: 10.1080/15325008.2013.862319
To link to this article: http://dx.doi.org/10.1080/15325008.2013.862319

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the Content) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions

Electric Power Components and Systems, 42(34):348360, 2014

C Taylor & Francis Group, LLC
Copyright 
ISSN: 1532-5008 print / 1532-5016 online
DOI: 10.1080/15325008.2013.862319

A Dual-loop Model Predictive Voltage

Control/Sliding-mode Current Control for Voltage
Source Inverter Operation in Smart Microgrids
Khaled M. Abo-Al-Ez,1,2 Ahmed Elaiw,2,3 and Xiaohua Xia2
1

Department of Electrical Engineering, Faculty of Engineering, Mansoura University, Mansoura, Egypt

Centre of New Energy Systems, Department of Electrical, Electronic and Computer Engineering, University of Pretoria,
South Africa
3
Department of Mathematics, Faculty of Science, King Abdulaziz University, Jeddah, Saudi Arabia

Downloaded by [Sabanci University] at 01:11 13 August 2015

CONTENTS
1. Introduction
2. State-Space Modeling of Three-Phase Inverter of the
Microgrid
3. Design of the Microgrid Controller
4. Performance Evaluation
5. Conclusion
References

AbstractThe design of a robust controller for the voltage source inverter is essential for reliable operation of distributed energy resources
in future smart microgrids. The design problem is challenging in the
case of autonomous operation subsequent to an islanding situation.
In this article, a dual-loop controller is proposed for voltage source
inverter control. The outer loop is designed for microgrid voltage and
frequency regulation based on the model predictive control strategy.
This outer loop generates reference inverter currents for the inner
loop. The inner loop is designed using a sliding-mode control strategy, and it generates the pulse-width modulation voltage commands
to regulate the inverter currents. A standard space vector algorithm
is used to realize the pulse-width modulation voltage commands.
Performance evaluation of the proposed controller is carried out for
different loading scenarios. It is shown that the proposed dual-loop
controller provides the specified performance characteristics of an
islanded microgrid with different loading conditions.

1.

Keywords: Smart microgrid, voltage source inverter, model predictive

voltage control, sliding-mode current control, robust control, smart grid
Received 31 December 2012; accepted 8 October 2013
Address correspondence to Dr. Ahmed Elaiw, Centre of New Energy
Systems, Department of Electrical, Electronic, and Computer Engineering,
University of Pretoria, Pretoria, 0002, South Africa. E-mail: a m elaiw@
yahoo.com
Color versions of one or more of the figures in the article can be found online
at www.tandfonline.com/uemp.

348

INTRODUCTION

There has been a great need to fulfill increasing demands in the

energy sector for the last decade. The need for sustainable and
reliable electricity supply requires active power distribution
networks. This promotes the possibility of bidirectional power
flow between the main grid and the consumers. Therefore, the
design of future distribution grids is moving away from traditional radial systems toward more networked grids [1]. In
this scenario, the self-controlled smart microgrid (SMG) concept is a promising approach. The smart grid concept has been
introduced to overcome problems of the power systems and
to achieve desired functionalities, such as self-healing, high
reliability, high power quality, and accommodation of a wide
variety of distributed generation (DG) and storage options [2].
To prevent power system mal-operation and instability, the process of the self-healing procedure should be as fast as possible.

Downloaded by [Sabanci University] at 01:11 13 August 2015

Abo-Al-Ez et al.: A Dual-loop Model Predictive Voltage Control/Sliding-mode Current Control for VSI Operation in Smart Microgrids

This is not achievable unless new progress in computational

modeling and optimization methods is used [3].
In the new energy generation and consumption model, a
key role is played by the SMG, i.e., clusters of distributed energy resources (DERs) supporting local loads for which energy
deficiencies or production excesses are solved by exchanging
power with the utility. DERs provide the advantage to users of
exploitation of renewable sources thus reducing the production costs [4]. The SMG permits several intelligent functions
that ensure consumer power reliability and optimize the use
of energy, mitigate environmental and power quality related
impacts, and manage the energy prices and overall costs [5].
SMGs contain smart energy efficient end-use devices,
DERs, advanced whole-building control systems, and integrated communications architecture [6]. Most modern DERs,
such as microturbines, photovoltaic systems, fuel cells, etc., are
load/grid interfaced through power electronic voltage source
inverters (VSIs) [7]. According to [8], the SMG concept assumes a cluster of loads and micro-sources, operating as a
single controllable system, where the inverter plays a key role
for a SMG to participate in energy management and voltage
and frequency control.
The SMG is managed by controlling the VSI interface of
the DERs to work either as islanded or synchronized to the
main electric grid. Upon separation from the main grid, the
current operation standard prohibits the autonomous operation of the microgrid due to safety issues [9]. In [8], it was
stated that Upon separation from a utility system, if a smart
grid DG stays stable, the inverter of the DG system regulates its
frequency. If there is only one inverter in the DG system, load
frequency control is accomplished through inverter reference
frequency using the PWM technique (p. 20). The management and control for the SMG are high non-linear problems
that are not only related to the environmental/economic optimization problem but also to frequency/voltage security and
dynamic stabilization in multi-mode operation [10].
The main challenge for islanded microgrid operation is to
provide the VSI with a controller that regulates voltage and
frequency despite load variation; droop control is widely used,
as reported in [1115]. In this control approach, the frequency
and voltage of each DER unit is a linear function of active
power and reactive power respectively. A major drawback of
this approach is that it is intended for a fixed load and cannot
accommodate large variations in the load parameters. To overcome this drawback, the control design uses one of the microgrid inverters to operate with voltage-frequency (V-f) control
to stabilize the microgrid voltage and frequency subsequent to
islanding situation. When the DC architecture is adopted, only
one inverter is used with the required controller. In the case
of AC architecture, as adopted in [16], one of the microgrid

349

VSIs is controlled in V-f control mode, while other inverters

are controlled for active/reactive power (P-Q) control.
DC microgrids are gaining high momentum under the smart
grid environment. In a DC microgrid system, a bidirectional
pulse-width modulated (PWM) AC-DC VSI is used to interface the DC microgrid to an AC system. Within the DC microgrid, DG units, such as photovoltaic cells, microturbines, and
wind turbines, are integrated. Storage units are also available
to provide energy backup for critical loads [17].
Some recent research addressed the issue of inverter control in SMGs [1821]. In [18], a microgrid VSI control scheme
was proposed based on H optimal control method to provide
a robust performance in the presence of the variation of power
factor correction capacitors. Although the H approach is
effective for providing robust stability, it may not guarantee
voltage tracking capabilities under loading parameters uncertainty. In [19], a repetitive controller of a VSI in a microgrid
was designed. This controller is difficult to stabilize for all unknown load disturbances and cannot obtain a very fast response
for a fluctuating load. In [21], a robust servomechanism problem (RSP) control was presented for autonomous operation of
a single-DG islanded microgrid with a structurally uncertain
plant. The parametric uncertainty is due to load variation about
their rated values. The design an uses only RSP controller for
voltage and frequency control of the microgrid. In [20], RSP
control is implemented by the internal model principle. This
controller is used for the outer voltage loop of a microgrid VSI
showing robust performance.
A discrete-time sliding-mode current controller is used as
an inner control loop to the RSP voltage controller for the purposes of overload protection. In this method, a current-limiting
function is employed to limit the inverter current commands
generated by the RSP voltage controller. This limiting function is not included in the design of an RSP voltage controller.
Large overshoot can occur when not using a current limiter,
which damages the inverter device. Moreover, some authors
[22] do not appreciate the complexity of the method and consider that the main problem of the RSP controller is the need
for relatively hard design (p. 465).
The hard design of the RSP controller method was discussed in [23] for studying the properties of a closed-loop
sampled data system for the RSP. It was shown that fundamental limitations exist on the closed-loop performance of the
RSP for a minimum phase sampled system, which are independent of the order of the plant, and of the transmission zero
structure of the continuous plant model. It was also shown
that fundamental limitations exist on the closed-loop performance of the RSP for a non-minimum phase sampled system
in the case of tracking and regulation due to disturbances in the
output.

Downloaded by [Sabanci University] at 01:11 13 August 2015

350

Electric Power Components and Systems, Vol. 42 (2014), No. 34

To overcome the load variation and the hard design problems, the design of model predictive control (MPC) is considered for voltage and frequency regulation in an islanded
microgrid. The advantages of the MPC over the RSP include
its facility of handling constraints and being able to use simple
models; it presents good tracking performance that is necessary in industrial applications, offers closed-loop stability,
and has inherent robustness [24]. In recent decades, MPC has
become one of the most widely used multi-variable control
algorithms in various industries, including chemical engineering, food processing, automotive, aerospace applications [25],
and recently in power systems [2632]. In MPC, a finite horizon optimal control problem is repeatedly solved, and the
input applied to the system is based on the obtained optimal open-loop control [24]. Recently, an unconstrained MPC
approach was presented for the voltage regulation of an islanded inverter of a DG unit [22]. In [22], the constraint on
the currents of the inverter was neglected; further, load parameters were incorporated in the control procedure, which
is impractical and requires real-time update of the model and
control design parameters. In [22], the state-space model of
the microgrid did not consider the parameters of the transformer that is used to connect the microgrid to the main
grid.
In the present article, a self-controlled microgrid is investigated. The coupling VSI is equipped with a smart controller
that can accommodate the islanded mode of operation despite
change in loading conditions. The issue of smart voltage and
frequency control is addressed to ensure stable operation for
consumer loads on one hand and the different renewable generation and storage systems connected on a common DC bus
on the other hand. This article proposes a robust dual-loop
model predictive voltage control/sliding-mode current control
strategy to regulate the voltage and frequency of an islanded
microgrid with different loading conditions. The inner loop
is designed using a sliding-mode control to limit the inverter
current under overload conditions. The outer loop is designed
using a constrained MPC to regulate the microgrid load voltages to their reference values. The benefit of this design is that
it takes full advantage of the fast response in controlling the
inverter current by sliding mode [20] on the one hand, it makes
use of the inherent capability of handling constraints of MPC
on the other. This article is organized as follows. Section 2
describes the state-space mathematical modeling of the threephase inverter of the microgrid. The dual-loop controller detailed design for the microgrid island is carried out in Section 3.
Performance evaluation of the proposed controller is done
for various operation scenarios using MATLAB/Simulink in
Section 4. Finally, Section 5 concludes the article.

FIGURE 1. Three-phase inverter interface of DC architecture

microgrid.

2.

STATE-SPACE MODELING OF THREE-PHASE

INVERTER OF THE MICROGRID

The three-phase inverter system considered herein is a PWM

inverter used for DER units, as shown in Figure 1. A constant
DC source is used to model most typical types of DER
units for the purposes of controller design. This approximate
modeling is reasonable when a mix of DER units operates
with energy storage systems with a DC architecture. A typical
three-phase PWM voltage inverter is used, with an L-C
output filter (with inductance L inv and capacitance Cinv ) and
a delta-wye transformer (with resistance RT and inductance
L T ) that acts as both a voltage transformer and an electrical
isolation to the load. Small capacitors (denoted as Cload ) are
added at the load side to provide further harmonic filtering
and stabilization of the load voltage [8]. According to [8], it
is assumed that the PWM space vector and power converter
generate ideal voltages according to the commands given by
the control. This assumption is valid due to the relatively high
switching frequency of the PWM inverter [8]. The state-space
model of the microgrid system in a three-phase abc reference
frame is derived based on Kirchhoff s current and voltage
laws and is given by the following equations:

d V iabc
dt

d I iabc
dt

d V Loadabc
dt

d Isd abc
dt

1
1

I iabc
Tri Isd abc ,
3Cinv
3Cinv
1
1

=
V pwm abc
V iabc ,
L inv
L inv
1
1

=
Isd abc
I Loadabc ,
CLoad
CLoad
RT
1

=
Isd abc +
Tr v V iabc
LT
LT
1

V Loadabc ,
LT
=

(1)
(2)
(3)

(4)

Abo-Al-Ez et al.: A Dual-loop Model Predictive Voltage Control/Sliding-mode Current Control for VSI Operation in Smart Microgrids

Downloaded by [Sabanci University] at 01:11 13 August 2015

where V iabc , V Loadabc , and V pwm abc represent the inverter voltages, load voltages, and PWM voltages in the abc reference
frame, respectively. As a notational convention, a vector in

the x yz reference frame is denoted as V x yz = [ Vx Vy Vz ]T .

Vectors I iabc , I Loadabc , and Isd abc denote the inverter currents,
load currents, and transformer secondary side currents in
the abc reference frame, respectively. Matrices Tri and Tr v
denote the current and voltage transformation of the delta-wye
transformer, given by

1 2 1
0
0 1

Tri = tr .
1 2
0
, Tr v = tr . 1 0
,
1
2 1
1
0 1 0
where tr is the transformer turns ratio.
The design of the proposed control technique requires that
the dynamic equations of the system in the three-phase abc reference frame are transformed into the qd0 stationary reference
frame as follows [8]:

I dq0 = K s I abc ,
V dq0 = K s V abc ,

where V can be V i , V Load , V pwm ; and I can be I i , I Load ,

and I sd . The transformation matrix K s is given as

1
1
1

2
2

2
3
3
.
Ks =
0

3
2
2

1
1
1
2
2
2
Using the above voltage and current transformation equations,
the state-space equations of the three-phase inverter system in
a dq0-axis reference frame can be written as

d V iqd0
1
1

=
I iqd0
Triqd0 Isdqd0 , (5)
dt
3Cinv
3Cinv

d I iqd0
1
1

=
V pwm qd0
V iqd0 ,
(6)
dt
L inv
L inv

d V Loadqd0
1
1

Isdqd0
I Loadqd0 ,
(7)
=
dt
CLoad
CLoad

d Isdqd0
RT
1

=
Isdqd0 +
Tr vqd0 V iqd0
dt
LT
LT
1

V Loadqd0 ,
(8)
LT
where Triqd0 = K s Tri K s1 and Tr vqd0 = K s Tr v K s1 .
The zero reference components of the inverter voltages and
currents will always be zero and can be discarded for a balanced
three-wire system of the inverter and balanced loads. The zero
components of the load voltages, load currents, and secondary
transformer currents can be non-zero under unbalanced load

351

conditions; however, these components are non-controllable

and have no use for control [8].
In this case, the above equations in the dq0 can be simplified
in the dq frame as

d V iqd
1
1

=
I iqd
Triqd Isdqd ,
(9)
dt
3Cinv
3Cinv

d I iqd
1
1

V pwm qd
V iqd ,
(10)
=
dt
L inv
L inv

d V Loadqd
1
1

=
Isdqd
I Loadqd ,
(11)
dt
CLoad
CLoad

d Isdqd
1
1
RT

Isdqd +
Tr vqd V iqd
V Loadqd ,
=
dt
LT
LT
LT
(12)


1
3
3
Triqd = tr
,

2
3 1


1 3
1
.
Tr vqd = tr
2
3
1

The state variables of the system are V iqd , I iqd , V Loadqd , and

Isdqd ; the control input of the system is V pwm qd ; and I Loadqd

is treated as a disturbance.
The continuous-time model in Eqs. (9) and (12) is rewritten
as
x = As x + Bs u + E s d,

(13)

y = Cs x + Ds u,

(14)

where the state variable, input, output and disturbance are

x = [ Viq Vid Iiq Iid
u = [ V pwm q V pwm d ]T ,
d = [ I Loadq I Loadd ]T ,

VLoadq VLoadd Isdq Isdd ]T ,

y = [ VLoadq VLoadd ]T ,

respectively, and

1
022
I22

3Cinv

022
L I22
inv
As =

022
022

1
Tr vqd
022
LT

022

I22

L
Bs = inv

0
22

022

022
022
022
1
I22
LT

1
Triqd

3Cinv

022

1
I22

CLoad

RT
I22
LT

352

Electric Power Components and Systems, Vol. 42 (2014), No. 34

022

022

,
Es =

1
I22

CLoad

Cs = [ 022

022

I22

022 ],

022
Ds = [ 022 ].
In this article, the controller design is based on the nominal continuous-time model of the system shown in Eqs. (13)
and (14).

Downloaded by [Sabanci University] at 01:11 13 August 2015

3.

DESIGN OF THE MICROGRID CONTROLLER

The proposed control strategy uses a dual-loop control structure of the following controllers:
1. inner sliding-mode inverter currents control loop and
2. outer model predictive load voltages control loop.
The outer loop regulates the load voltages to follow the 50or 60-Hz balanced three-phase reference voltages, which are
decided by the microgrid operators. This voltage control loop
generates the inverter current set values, which are limited. The
inner current control loop generates the PWM gating voltages
to regulate the inverter currents to follow the inverter currents
set values [8]. Figure 2 shows the block diagram presenting
the control structure.
A standard voltage space vector algorithm is used to realize
the PWM command voltages applied to the VSI. In the next two
subsections, the design of the proposed dual-loop controller
is given. The discrete-time state-space equations are adopted
for the two-loop control design in the dq stationary reference
frame.
3.1.

Discrete-time Sliding-mode Current Controller

The discrete-time sliding-mode controller is used in the inner

current control loop to limit the inverter current under overload conditions because it provides fast and non-overshoot
response [8]. To design the discrete-time sliding-mode current

controller, the model of the three-phase inverter with an L-C

filter is considered without the transformer and load. The sec

ondary transformer current Isdqd is treated as a disturbance to

the current controller. The state equations describing the system used for controller design are Eqs. (9) and (10). The statespace representation is as follows:
(15)
x1 = A1 x1 + B1 u 1 + E 1 d1 ,
y1 = C1 x1 + D1 u 1 ,
(16)

1
.I22
022
022

3Cinv

A1 =
,
, B1 = 1
1
.I22
.I22
022
L inv
L

inv
1
3C .Triqd
E 1 = inv
, C1 = [ 022 I22 ], D1 = [022 ],
022
and the internal state, input, outputs and disturbance are
x1 = [ Viq

Vid

y1 = [ Iiq

Ii d ] T ,

Ii q

Ii d ] T ,

u 1 = [ V pwm q

d1 = [ Isdq

V pwm d ]T ,

Isdd ]T .

The state-space equations (Eqs. (15) and (16)) are transformed

to a discrete form as follows:
x1 (k + 1) = A1 x1 (k) + B1 u 1 (k) + E 1 d1 (k),
y1 (k) =

C1 x1 (k)

D1 u 1 (k),

(17)
(18)

where

Ts
e A1 (Ts ) B1 d,
A1 = e A1 Ts , B1 =
0

Ts

E1 =
e A1 (Ts ) E 1 d, C1 = C1 , D1 = D1 ,
0

and Ts is the sampling period, which, in this case, is equal to the

PWM period [8]. To force the inverter currents to follow their
set values (decided by the voltage controller), the sliding-mode
surface is chosen such that
S(k) = y1 (k) y1,r e f (k) = C1 x1 (k) y1,r e f (k)

= I iqd I iqd ,r e f .
Thus, the sliding surface S(k) is selected such that when the
discrete sliding-mode exists,

S(k) = 0 or I iqd = I iqd ,r e f .

The discrete-time sliding mode can be reached if control input
u 1 (k) is designed to be the solution of the following:
S(k + 1) = y1 (k + 1) y1,r e f (k + 1).

FIGURE 2. Proposed MPC/Sliding-mode controller structure.

The control law that satisfies the above equation is known as

equivalent control and can be derived as follows:

S(k + 1) = C1 x1 (k + 1) I iqd ,ref = 0.

Abo-Al-Ez et al.: A Dual-loop Model Predictive Voltage Control/Sliding-mode Current Control for VSI Operation in Smart Microgrids

Then, from Eq. (17), the equivalent control to be implemented

in the inner current control loop can be obtained as

u 1eq (k) = (C1 B1 )1 [ I iqd ,ref C1 A1 x1 (k) C1 E 1 d1 (k)].

(19)
3.2.

Discrete Time Model Predictive Voltage Controller

Downloaded by [Sabanci University] at 01:11 13 August 2015

Since the MPC is the outer loop, the plant seen by the voltage
controller is the combination of the true plant (three-phase
inverter with an L-C filter and the transformer) along with
the discrete-time sliding-mode current controller. To design the
voltage controller, the following discrete-time nominal model
of Eqs. (13) and (14) is used:
x(k + 1) = As x(k) + Bs u(k),

(20)

y(k) = Cs x(k) + Ds u(k),

(21)

where the state variable, input, and output are

x = [ Viq Vid Iiq Iid VLoadq VLoadd Isdq
u = [ V pwm q
As = e As Ts ,

Isdd ]T ,

V pwm d ]T , y = [ VLoadq VLoadd ]T ,

Ts
B1 =
e As (Ts ) Bs d, Cs = Cs Ds = Ds .

VLoad,ref,an (k) = Vm sin(2 f ref k Ts ),



2
,
VLoad,ref,bn (k) = Vm sin 2 f ref k Ts
3


2
,
VLoad,ref,cn (k) = Vm sin 2 f ref k Ts +
3
where Vm is the phase voltage magnitude, and f ref is the fundamental reference frequency.
The load voltage reference in the qd0 frame is given by

V Load,r e f,abc = [ VLoad,ref,an VLoad,ref,bn VLoad,ref,cn ]T ,

V Load,r e f,qd0 = K s V Load,r e f,abc

= [ VLoad,ref,q

VLoad,ref,d

x(k + 1) = As x(k) + Bs (C1 B1 )1

[ I iqd ,r e f C1 A1 x1 (k) C1 E 1 d1 (k)].

The Augmented Plant of the Voltage Controller

To augment the discrete-time sliding-mode controller in the

plant, the control law of the discrete-time sliding-mode current
controller u 1eq (k) is used to equal u(k) in Eqs. (20) and (21);

(22)

The inverter current set value is the input of the MPC of the
voltage; therefore u v (k) = [ Iiq ,r e f Iid ,r e f ]T . Then Eqs. (20)
and (21) become
xv (k + 1) = Av xv (k) + Bv u v (k),
yv (k) =
where

Cv xv (k),

(23)
(24)

1 

Av = As Bs C1 B1
C1 A1 F1 + C1 E 1 F2 ,

1
Bv = Bs C1 B1 , Cv = Cs ,


I22 022 022 022
,
F1 =
022 I22 022 022
F2 = [ 022

022

I22

022 ].

The next subsection uses the augmented plant to design the

model predictive voltage controller.
3.2.2.

Design of the Discrete-Time Model Predictive

Voltage Controller

MPC is essentially a class of computer control algorithms to

control the future behavior of a system through the use of
an explicit model of the system. At each control step, the
MPC algorithm computes an open-loop sequence of controls
to optimize the future system behavior. For the MPC design,
the system in Eqs. (23) and (24) is considered. Bounds on the
manipulated variables are assumed, as follows:
u max u v (k) u max ,
where u max is the maximum peak value of the inverter currents.
3.2.3.

Augmented Model

Equations (15) and (16) are first transformed into an increment

model that is appropriate for MPC design purposes [33]:
xv (k + 1) xv (k) = Av (xv (k) xv (k 1))
+ Bv (u v (k) u v (k 1)),
yv (k + 1) yv (k) = Cv (xv (k + 1) xv (k)),

VLoad,ref,0 ]T .

Again, only the qd componenets are needed for the MPC

design:

V Load,r e f,qd = [ VLoad,ref,q VLoad,ref,d ]T .

3.2.1.

so, with substitution in those equations;

The load voltage reference to be tracked by the MPC is generated in the abc reference frame then transformed to the qd
needed for the controller. The referene phase to neutral load
voltages in the abc frame is given by

353

or
xv (k + 1) = Av xv (k) + Bv u v (k),
yv (k + 1) = yv (k) + Cv Av xv (k) + Cv Bv u v (k),
and in final form with integral action:




xv (k + 1)
Av
xv (k)
0
=
yv (k + 1)
Cv Av I
yv (k)

Bv
[u v (k)] .
+
Cv Bv

(25)

Electric Power Components and Systems, Vol. 42 (2014), No. 34

354

By choosing a new state variable vector x(k) = (xv (k),

yv (k))T , Eq. (25) the becomes
x(k + 1) = Ax(k) + Bu(k),
y(k + 1) = C x(k),
where

A=

Downloaded by [Sabanci University] at 01:11 13 August 2015

3.2.4.

Av

Cv Av


,

B=

(26)
(27)

Bv
Cv Bv

C = [0I ], u = u v ,

Prediction of State and Output

Based on the augmented model in Eqs. (26) and (27), the predicted plant output is calculated using control input as the adjustable variable. Suppose that at sampling instant k the output
measurement y(k) is available; the future input trajectory is denoted by u(k|k), u(k + 1|k), . . . , u(k + Nc 1|k), where
Nc is the control horizon.
The predicted state and output over a prediction horizon N p
are given by
x(k + 1|k), x(k + 2|k), . . . , x(k + N P |k),
y(k + 1|k), y(k + 2|k), . . . , y(k + N P |k),
where x(k + 1|k) and y(k + 1|k) represent the future state and
output, respectively, at sampling instant k + j using initial
x(k|k) = x(k). It is assumed that Nc N p and u(k + Nc +
i|k) = 0 (i = 0, 1, . . . , N p Nc 1).
It is noted that the predicted state and output can be represented as
x(k + j|k) = A x(k) +
j

j1


A ji1 Bu(k + i|k),

i=0
j1

y(k + j|k) = C A x(k) +

j

CA

ji1

Bu(k + i|k).

i=0

3.2.5.

Denote the following vector U (k) = [u(k|k), u(k +

1|k), . . . , u(k + Nc 1|k)]T . An objective function resulting in the control inputs to penalize deviations of the predicted
control outputs is first defined from a given reference trajectory
r (k + i|k):
V (k) =

satisfactory dynamic performance. The optimization problem

is subject to the following constraints:
u max u(k + i|k) u max ,

Optimization

Np


FIGURE 3. Load voltage and current with reference voltage

variation.

It is noted that objective function V (k), as well as the input

constraint, can be rewritten in terms of U (k) [33]. This leads
to the following standard optimization problem:
1
min U (k)T U (k) + T U (k),
U (k) 2
subject to

y(k + i|k) r (k +

i=0
N
c 1


i|k)2Q(i)

u(k + i|k)2R(i) ,

i=0

where Q(i) and R(i) are the weights. Parameters Nc , N p , Q(i),

and R(i) are all design parameters that can be tuned to give

i = 0, 1, . . . , Nc 1.

U (k) ,

where , , , and are matrices of specific type; see [33].

This optimization problem can be solved by quadratic programming. Let the solution of the optimization problem be
denoted by

U (k) = [u (k|k), u (k + 1|k), . . . , u (k + Nc 1|k)]T .

Abo-Al-Ez et al.: A Dual-loop Model Predictive Voltage Control/Sliding-mode Current Control for VSI Operation in Smart Microgrids

3.2.6.

MPC Algorithm

The idea of MPC is based on the receding horizon control

principle, which can be summarized as follows:
(i) at sampling time instant k, solve the optimization problem using the current measurement y(k);
(ii) the first element of the resulting optimal controller,
i.e., u (k|k), is applied to the plant in the interval
[k, k + 1] (the remaining {u (k + 1|k), . . . , u (k +
Nc 1|k} is disregarded;

Downloaded by [Sabanci University] at 01:11 13 August 2015

(iii) the procedure based on the output measurement y(k +

i|k) is repeated from step (i) for the next sampling
instants i + k(i = 1, 2, . . .).
4.

PERFORMANCE EVALUATION

The load voltage reference to be traked by the MPC is given

in RMS value, then generated in the three-phase abc reference

FIGURE 4. Load voltage and current at transition from reference voltage 90 V (RMS) to reference voltage 120 V (RMS)
of Figure 3.

355

frame, then transformed to the qd0 stationary reference frame

needed for the controller. It is therefore required for different loading conditions to measure load voltage and current
in the abc reference frame with the required reference frequency of 60 Hz and phase angle difference of 120 between
the three phases. The measurements are also shown in the qd0
stationary reference to further explain the effectiveness of the
controller by observing the qd0 load voltage and current, especially the zero-sequence value in different loading conditions.
The tests focus on the stability of the controller to regulate the
voltage and frequency in cases of microgrid reference voltage
variation, load variation, single- and two-phase operation, and
unbalanced loading.
The performance evaluation is intended to show the effectiveness of the controller design under different operation
scenarios. The controller is designed based on the nominal
system of Eqs. (13) and (14) and is applied to the plant that
presents the three-phase inverter, filter, and loads described by

FIGURE 5. Load voltage and current with load resistance

variation.

356

Electric Power Components and Systems, Vol. 42 (2014), No. 34

Value

Quantity
VSI filter inductance L inv
VSI filter capacitance Cinv
Load side capacitors CLoad
Transformer inductance L T
Transformer resistance RT
Transformer turns ratio tr
System nominal frequency f 0
DC bus voltage VDC

298 H
540 F
90 F
0.03 p.u.
0.03 p.u.
120/245
60 Hz
540 V

Downloaded by [Sabanci University] at 01:11 13 August 2015

TABLE 1. Parameters of the system in Figure 1

Eqs. (5) and (8) with I Loadabc = V Loadabc /Rload . Here Rload
is an uncertain load parameter.
The parameters of the test system shown in Figure 1 are
given in Table 1.
For performance evaluation and verification of the proposed
controller, the following tests are performed.

FIGURE 6. Load voltage and current at transition of load

resistance variation of Figure 5.

Test 1: Maintain the magnitude and the frequency of the

island voltage at desired reference voltage levels.
In this test, the qd-reference voltage (RMS) value is step
changed from 90 to 120 V, then back to 90 V, with a load
resistance of 20 ; the simulation is done for 1 sec, as
shown in Figure 3. Transient from 90 to 120 V (RMS)
is shown in Figure 4. The results show that the controller
is capable of tracking the reference voltage magnitude and
frequency with zero steady-state error. The load and inverter
line currents are increased to accommodate the increased
demand of the load due to microgrid voltage increment.
Test 2: Maintain the stability of the islanded microgrid in
the presence of load uncertainties, which are presented
by load resistance variations.
Load resistance is step decreased from open circuit to 80,
40, and 20 , and finally to open-circuit case gain, as shown
in Figure 5. As the load resistance changes to smaller value,
the VSI has to increase its real current to compensate the
increase of real power of the load. Step changes show the effectiveness of the proposed controller in dealing with abrupt

FIGURE 7. Load voltage and current with single-phase load

operation.

Downloaded by [Sabanci University] at 01:11 13 August 2015

Abo-Al-Ez et al.: A Dual-loop Model Predictive Voltage Control/Sliding-mode Current Control for VSI Operation in Smart Microgrids

357

load variations. Transient from 80 to 40  is shown in Figure 6.

Test 3: Maintain the voltage stability of the islanded microgrid with single- and two-phase load operation.
The simulation for both cases is performed for 0.1 sec. To
simulate single-phase operation, the resistance of phase a
is set at 20 , while phases b and c are open circuited. The
simulation results are presented in Figure 7. It is observed
that the controller adjusts the microgrid load voltage at
120 V (RMS), while the line current is drawn by phase a
only.
When testing two-phase operation, phase a and b resistances are set to 20 , while phase c is open circuited.
The simulation results are presented in Figure 8. It is observed that the controller adjusts the microgrid load voltage
at 120 V (RMS), while line currents are drawn by phase a
and b.
Test 4: Maintain the voltage stability of the islanded microgrid for unbalanced loading.

Assuming a full load resistance of 20 , phase a is 50%

loaded, phase b is 80% loaded, and phase c is 70% loaded.
The simulation is performed for 0.1 sec. The results are
presented in Figure 9.
Test 5: Transient performance for transition from open
circuit to short circuit at the output terminals using the
proposed controller.
The final verification to the proposed control algorithm is
for the ability to limit the inverter currents under short
circuits at the output terminal, as shown in Figs. 10
and 11.

FIGURE 8. Load voltage and current two-phase load operation.

FIGURE 9. Load voltage and current with unbalanced loading.

(a) With constrained MPC voltage controller: The proposed MPC voltage controller provides the reference
inverter currents, which are constrained to prevent the
inverter currents from exceeding the rated values. In
this case, there is no need for a separate overload current limiting function as it is embedded in the MPC
design. Figure 10 shows the system response to the
transition from open circuit to short circuit.

Downloaded by [Sabanci University] at 01:11 13 August 2015

358

Electric Power Components and Systems, Vol. 42 (2014), No. 34

FIGURE 10. Transition from open circuit to short circuit.

(b) With RSP voltage controller: This test is performed

using the RSP for voltage controll as proposed in [20].
In this method, the current limiting function is performed to limit the inverter current commands generated by the RSP voltage controller. This limiting
function is not included in the RSP voltage controller
design, and is accomplished by limiting the magnitude
of the inverter current commands vector to 300% of
the rated values and maintaining its original direction.
Figure 11 shows the response to the transition from
open circuit to short circuit, mainly showing the response of the inverter currents with and without the
current limiting function. Large overshoot is observed
in the case of not using a current limiter, which is
likely to damage the inverter device.
The simulation results of Tests 14 show that, in cases of
a balanced three-phase load and single-phase load operations,
the zero component of the load current in the qd0 frame is
always zero, while in cases of an unbalanced three-phase load
and two-phase load operations, it is non-zero. It is also ob-

FIGURE 11. Transition from open circuit to short circuit in

case of using RSP control.

served that the zero component of the load voltage in the qd0
frame is always zero for all loading conditions.
5.

CONCLUSION

A proposed dual-loop controller for the V-f control of a VSI in

an islanded microgrid is presented in this article. The controller
design is based on employing a cascaded control strategy of
two loops. The inner loop is a fast response current controller
designed using sliding-mode control to regulate the VSI current under overload conditions. The outer loop is a voltage
controller designed using MPC to regulate the microgrid voltage and frequency. A dynamic model of a test system and
controller is developed. The performance evaluation of the
proposed control strategy is tested under different operating
conditions, showing the robustness of the proposed strategy
for autonomous operation of SMGs.
REFERENCES
[1] Strzelecki, R., and Benysek, G., Power Electronics in Smart
Electrical Energy Networks, London: Springer-Verlag, Chap.
1., pp. 15, 2008.

Downloaded by [Sabanci University] at 01:11 13 August 2015

Abo-Al-Ez et al.: A Dual-loop Model Predictive Voltage Control/Sliding-mode Current Control for VSI Operation in Smart Microgrids
[2] Moaddabi, N., Hosseinian, S. H., and Gharehpetian, G. B.,
Practical framework for self-healing of smart grids in stable/unstable power swing conditions, Elect. Power Compon.
Syst., Vol. 40, No. 6, pp. 575596, 2012.
[3] Moaddabi, N., and Gharehpetian, G. B., Wide-area method
for self-healing of smart grids in unstable oscillations, Elect.
Power Compon. Syst., Vol. 41, No. 4, pp. 365382, 2013.
[4] Erseghe, T., and Tomasin, S., Power flow optimization for
smart microgrids by SDP relaxation on linear networks, IEEE
Trans. Smart Grid, Vol. 4, No. 2, pp. 751762, 2013.
[5] Girgis, A. A., and Makram, E. B., Innovations in power systems, Proceedings of 14th Int. Middle East Power Systems
Conference (IEEE MEPCON10), Cairo, Egypt, 1921, December 2010.
[6] Gellings, C. W., The Smart Grid: Enabling Energy Efficiency
and Demand Response, Liburn, GA: Fairmont Press, Chap. 1,
pp. 15, 2009.
[7] Dugan, R. C., McGranaghan, M. F., Santos, S., and Wayne
Beaty, H. , Electrical Power Systems Quality, 2nd ed. New
York: McGraw Hill, Chap. 9, pp. 386388, 2004.
[8] Keyhani, A., Marwali, M., and Dai, M., Integration of Green
and Renewable Energy in Electric Power Systems, Hoboken,
NJ: John Wiley & Sons, Chap. 2, pp. 2630, 2010.
[9] IEEE, IEEE Standard for interconnecting distributed resources
with electric power systems, IEEE Standard 1547-2003, pp.
116, July 2003.
[10] Dou, C. X., Liu, D. L., Jia, X. B., and Zhao, F., Management
and control for smart microgrid based on hybrid control theory, Elect. Power Compon. Syst., Vol. 39, No. 8, pp. 813832,
2011.
[11] Piagi, P., and Lasseter, R. H., Autonomous control of microgrids, Proceedings of IEEE Power Engineering Society General Meeting, 1822, Montreal, June 2006.
[12] Lasseter, R. H., and Piagi, P., Control and Design of Microgrid Components, Tempe, AZ: PSERC Publication, Chap. 3,
pp.1926, 2006.
[13] Nikkhajoei, H., and Lasseter, R. H., Distributed generation
interface to the CERTS microgrid, IEEE Trans. Power Deliv.,
Vol. 24, No. 3, pp. 15981608, 2009.
[14] Sao, C. K., and Lehn, P. W., Autonomous load sharing of
voltage source converters, IEEE Trans. Power Deliv., Vol. 20,
No. 2, pp. 10091016, 2005.
[15] Yu, X., Khambadkone, A. M., Wang, H., and Terence, S. T. S.,
Control of parallel-connected power converters for low-voltage
microgridPart I: A hybrid control architecture, IEEE Trans.
Power Electron., Vol. 25, No. 12, pp. 29622970, 2010.
[16] Abo-Al-Ez, K. M., Xia, X., and Zhang, J., Smart interconnection of a PV/wind DG micro grid with the utility distribution
network, Proceedings of Industrial and Commercial Use of
Energy Conference (ICUE), Cape Town, South Africa, 14-17
August 2012.
[17] Ahmed, A., Radwan, A., and Abdel-Rady Mohamed, Y., Linear active stabilization of converter-dominated DC microgrids,
IEEE Trans. Smart Grid, Vol. 3, No. 1, pp. 203216, 2012.
[18] Li, Y., Vilathgamuwa, D. M., and Loh, P. C., Robust control
scheme for a microgrid with PFC capacitor connected, Proc.
IEEE Industry Applications Conference, 40th IAS Annu. Meeting, Vol. 4, pp. 24412448, 26 October, 2005.

359

[19] Marwali, M., and Keyhani, A., Control of distributed generation systemspart I: Voltages and current control, IEEE Trans.
Power Electron., Vol. 19, No. 6, pp. 15411550, 2004.
[20] Marwali, M., Dai, M., and Keyhani, A., Robust stability analysis of voltage and current control for distributed generation
systems, IEEE Trans. Energy Convers., Vol. 21, No. 2, pp.
516526, 2006.
[21] Karimi, H., Davison, E., and Irvani, R., Multivariable
servomechanism controller for autonomous operation of
distributed generation unit: Design and performance evaluation, IEEE Trans. Power Syst., Vol. 25, No. 2, pp. 853865,
2010.
[22] Bahrani, B., and Ruler, A., Model predictive-based voltage
regulation of an islanded distributed generation unit, IEEE
Conference on Energy Conversion Congress and Exposition
(ECCE), pp. 465472, Phoenix, AZ, 1722 September 2011.
[23] Gemma, L., Performance Limitations in the Robust Servomechanism Control Problem for Discrete and Sampled Data System,
Ph.D. Thesis, University of Toronto, December 2000.
[24] Mayne, D. Q., Rawlings, J. B., Rao, C. V., and Scokaert, P.
O. M., Constrained model predictive control: Stability and
optimality, Automatica, Vol. 36, No. 6, pp. 789814, 2000.
[25] Qin, S. J., and Badgwell, T., A survey of industrial model
predictive control technology, Control Eng. Pract., Vol. 11,
No. 7, pp. 733764, 2003.
[26] Otomega, B., Marinakis, A., Glavic, M., and Van Cutsem, T.,
Model predictive control to alleviate thermal overloads, IEEE
Trans. Power Syst., Vol. 22, No. 3, pp. 13841385, 2007.
[27] Ekpenyong, U. E., Zhang, J., and Xia, X., An improved robust model for generator maintenance scheduling, Elect. Power
Syst. Res., Vol. 92, pp. 2936, 2012.
[28] Xia, X., Zhang, J., and Elaiw, A. M., An application of model
predictive control to the dynamic economic dispatch of power
generation, Control Eng. Pract., Vol. 19, pp. 638648, 2011.
[29] Elaiw, A. M., Xia, X., and Shehata, A. M., Application of
model predictive control to optimal dynamic dispatch of generation with emission limitations, Elect. Power Syst. Res., Vol.
84, pp. 3144, 2012.
[30] Elaiw, A. M., Xia, X., and Shehata, A. M., Minimization of
fuel costs and gaseous emissions of electric power generation by
model predictive control, Math. Prob. Eng., Vol. 3013, Article
ID 906958, 2013.
[31] Zhang, J., and Xia, X., A model predictive control approach
to the periodic implementation of the solutions of the optimal
dynamic resource allocation problem, Automatica, Vol. 47, pp.
358362, 2011.
[32] van Staden, A. J., Zhang, J., and Xia, X., A model predictive
control strategy for load shifting in a water pumping scheme
with maximum demand charges, Appl. Energy, Vol. 88 pp.
47854794, 2011.
[33] Maciejowski, J. M., Predictive Control with Constraints,
Harlow, England: Prentice Hall, Chap. 3, pp. 7390, 2002.

BIOGRAPHIES
Khaled M. Abo-Al-Ez received the B.S., M.S., and Ph.D.
degrees from the Department of Electrical Engineering,

360

Electric Power Components and Systems, Vol. 42 (2014), No. 34

Downloaded by [Sabanci University] at 01:11 13 August 2015

Mansoura University, Mansoura, Egypt, in 2000, 2003, and

2011, respectively. He was a visiting scholar at the University of New Brunswick, Fredericton, Canada from Sept. 2005
to Aug. 2006. He was a post doctoral fellow at the Centre
of New Energy Systems (CNES) in the University of Pretoria, Pretoria, South Africa from Mar. 2012 to Feb. 2013.
Dr. Abo-Al-Ez is currently an assistant professor of electrical
power engineering in Mansoura University, Mansoura, Egypt.
His research interests are renewable energy generation systems, decentralized generation, mathematical and computer
modeling, power system optimization, power system planning
and automation, power electronics, power quality and smart
grids.
Ahmed Elaiw received his Ph.D. from Budapest University of
Technology and Economics, Budapest, Hungary. He was a post
doctoral fellow at the Centre of New Energy Systems (CNES)
in the University of Pretoria, Pretoria, South Africa from Aug.
2007 to Jul. 2008. Also, he was a visiting professor at CNES,
University of Pretoria from Jun. 2012 to Aug. 2012. Currently,

he is an associate professor of Mathematics at King Abdulaziz

University, Saudi Arabia. His research interests include control
systems, power system optimization and smart grids.
Xiaohua Xia is a professor in the Electrical, Electronic and
Computer Engineering at the University of Pretoria, South
Africa, director of the Centre of New Energy Systems, and the
director of the National Hub for the Postgraduate Programme
in Energy Efficiency and Demand-side Management. Prof. Xia
is a fellow of the Institute for Electronic and Electrical Engineers (IEEE), a fellow of the South African Academy of Engineering (SAAE), and a member of the Academy of Science
of South Africa (ASSAf). He has an A rating from the South
African National Research Foundation (NRF). He has been an
associate editor of Automatica, IEEE Transactions on Circuits
and Systems II, IEEE Transactions on Automatic Control, and
specialist editor (control) of the SAIEE Africa Research Journal. His research interests are control systems and automation,
and more recently, the modeling and optimization of energy
systems.